CN1442032A - High expandable internet super server - Google Patents
High expandable internet super server Download PDFInfo
- Publication number
- CN1442032A CN1442032A CN01812793A CN01812793A CN1442032A CN 1442032 A CN1442032 A CN 1442032A CN 01812793 A CN01812793 A CN 01812793A CN 01812793 A CN01812793 A CN 01812793A CN 1442032 A CN1442032 A CN 1442032A
- Authority
- CN
- China
- Prior art keywords
- mainboard
- server
- connector
- blade
- motherboard
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
-
- H—ELECTRICITY
- H05—ELECTRIC TECHNIQUES NOT OTHERWISE PROVIDED FOR
- H05K—PRINTED CIRCUITS; CASINGS OR CONSTRUCTIONAL DETAILS OF ELECTRIC APPARATUS; MANUFACTURE OF ASSEMBLAGES OF ELECTRICAL COMPONENTS
- H05K7/00—Constructional details common to different types of electric apparatus
- H05K7/14—Mounting supporting structure in casing or on frame or rack
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/16—Constructional details or arrangements
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Human Computer Interaction (AREA)
- Cooling Or The Like Of Electrical Apparatus (AREA)
- Mounting Of Printed Circuit Boards And The Like (AREA)
- Steroid Compounds (AREA)
Abstract
A scalable Internet engine (100) is comprised of a large number of commercially available server boards (134) each arranged as an engine blade (132) in a power and space efficient cabinet (110). The engine blades (132) are removably positioned in a front side (112) of the cabinet (110) in a vertical orietation. A through plane (130) in the middle of a chassis assembly within the cabinet provides common power and control peripheral signals to all engine blades. I/O signals for each engine blade are routed through apertures in the through plane to interface cards positioned in the rear of the cabinet. The scalable engine (100) can accommodate different types of server boards in the same chassis assembly because of a common blade carrier structure. Different types of commercially available motherboards (200) are mounted in the common blade carrier structure that provides a uniform mechanical interface to the chassis assembly. A specially designed PCI host board that can plug into various types of motherboards has a first connector (205) for connecting to the through plane and second connector for connecting to the interface cards.
Description
Technical field
In a broad sense, the present invention relates to computer programming system.Say to narrow sense, the high expandable internet super server that the present invention relates to is made of a large amount of commercial server mainboards, and these mainboards exist as superserver blade mainboard (scalable server engine) in the assembling framework with enough electric power and space.
Background technology
Commercialization ISP and server service provider appear at the fast development that has promoted the Internet to a great extent, for example Internet service provider (ISPs), ASP (ASPs), stand alone software merchant (ISVs), the scheme consulting developer (ESPs) of enterprise and Management Advisory Services developer (MSPs) or the like.About above these services, here not clear and definite definition, but as a rule, the service that these service suppliers and main process equipment merchant are provided will be catered to exactly, most of even whole clients' demand, these demands are mode then about host application, website exploitation, managing eBusiness and server allotment with earning construction cost or periodic service expense.For example, for the server allotment, expense mainly results from according on client's special demands and the hardware and software specification for its application and website appointment setting.As purpose of the present invention, term " host services " intention is provided by the various dissimilar service that is provided by service supplier in this field and host services merchant.For simplicity, we claim these service suppliers and main process equipment merchant for " server service provider " (HSP).
Just as providing the mode of line, commercial HSPs to offer one of user by international telephony network, telephone operator can enter the passage that network host is used between their client.HSPs is used to provide the computer equipment of host application and service to be commonly referred to server.In the simplest mode, this server can be one and be connected to PC on the Internet by socket that it can move according to the custom-designed special software of the customer requirements of this server.Employed various mode when providing host services for HSPs, most of HSPs will use one group of server set that connects with internal network.This server set is exactly our usually said " server farm " (server farm).In this " group ", each server can be finished its unique task, also can share multinomial different task by several servers, for example mailbox server, the webserver, certificate server and can take into account management server.For example when providing host services for the World Wide Web website, the single webserver is generally a lot of small-sized World Wide Web websites gatherings and provides support, and big website then need be supported operation by special server.General background knowledge about Internet and host services supplier sees also Geoff huston,
ISP Survival Guide: manage the strategy of ISP in the competitive environment, (1999).
Along with the increase day by day to the Internet service demand, the market space of Internet industry is also just increasing.A kind of method that satisfies this class market demand is exactly to utilize the computer system of bigger ability as server.Large-scale main frame and medium-sized computer system are used to do the server of large-scale internet sites and company's network, most HSPs is owing to the expense of considering great number, complexity and lack flexibility and be not inclined to the use large computer system, these HSPs are ready to use " server farm " (the server farm) that is made up of many PC servers to support operation on the contrary, these servers are connected on the shared the Internet line or modulator-demodulator group, also can enter in one group of disc driver sometimes.When HSP increased a host services client, one or more PC servers manually was increased among the HSP " server farm ", and the client has installed specific software and data (for example, Web content) for this reason.In this way, the hardware that HSP only disposes certain level is supported its current client's demand, no less important, HSP can collect the early stage installation cost to pay the prime cost of this hardware to the client.By making in this way, the large computer system expense that HSP needn't the subsidiary bulk redundancy capacity of prepayment, and these expenses are can not be HSP generation income immediately.Compare for different clients provides support with mainframe of use, " server farm " provides a cover simple method, and this method can be guaranteed the fail safe and the data integrity of customer data in the running environment of different clients' coexistences.If the server that is a customer service only loads this client's software, and only be connected to this client's data, then customer information thereby independent process are protected safely.
Up to the present, HSPs has the mode of two kinds of foundation " server farm ".The one group of homogeneity personal computer system (hardware and software) that is provided by same manufacturer just is provided a kind of mode.The personal computer system that is provided by different manufacturers then is provided another kind of mode.The benefit that the homogeneity mode is brought is exactly that HSP only need support a single server platform, but it also makes HSP be limited in this single platform simultaneously, heterogeneous mode use the system from different manufacturers make its operation more flexibly and a conspicuous benefit be can use optimal server hardware and software platform at specific client and task, but to be exactly running environment complicated more and handle all troubles that the different server platform is brought for the cost that this flexibility is paid.
Along with the capacity increase of HSPs, the also corresponding increase of number of servers in machine room and data center.In order to make the space better utilization of machine room and data center, HSP will set up tier rack type rack (rack mounted cabinet) and place a large amount of servers.In this tier rack type rack, the PC server can vertically also can level be stacked with switchgear, modulator-demodulator.As the further improvement of conventional personal computer laying method, server develops into the cabinet-type layout and places, and just we are said as " pancake case " or " U " formula general layout, and like this, the one or more hole slot of rack is advanced in each server horizontal slip.The homogeneity mode server of layout comprises the RaQ server group of Cobalt Network System by this way, the A-class server group of the Netfinity server group of IBM and Hewlett-Packard.
Although utilize layer frame build-in services device certain progress to be arranged saving aspect the space to compare with traditional tower case PC server (tower) modes of emplacement, in server farm, install or increase new tier rack type server and remain a difficult task, comprise that a large amount of electric wires and circuit are plugged into and many networks and software installment work all must manually be finished.In addition, because the tier rack type server is designed to an independently computer system, each server all has the power supply of oneself.In order to keep the reasonability of server whole cost, these individual power supplys are not expensive so be relative poor efficiency yet.The negative results that " server farm " be made up of a large amount of tier rack type servers brought is exactly to consume a large amount of electricity and a large amount of heat energy of waste.This problem is allowing the maximum server successively to build in the ultra-thin 1U formula of putting more obvious.Because each 1U server is horizontal location in layer frame case, the smooth and height 1U server of air flows also makes the problem of utilizing an enough big fan to solve circulation of air become impossible in the case.
The method of a more effective placement server is the Chatterbox server ranks (server line) of ChatCom company.The frame of Chatterbox system (chassis) is supported front loading, the heat transfer model of some, comprises server board (motherboard), power supply and RAID driver.In this system, specially designed server board is just in time opposite with the server that vertical direction slides to mainboard frame and layer frame dress horizontal direction.The design of motherboard then can adapt to the different editions of the Intel processor of the main CPU of early stage conduct, this motherboard comprises a COM monitor port that is connecting modulator-demodulator, and microcontroller and the related software of a supervision modulator-demodulator and managing I/O port to the line of main CPU also is installed on this motherboard.Main CPU is Intel or Intel386,486, Pentium or Pentium Pro processor.Motherboard is not that the design of ChatCom company oneself is exactly common these the early stage servers of AT formula plate from early stage PC motherboard manufacturer, and electric power can directly enter motherboard from the heat exchange power supply the mainboard frame.With present server use standard A TX supply of electric power with set electrical management and controlled function different be that these early stage servers not have the electrical management and the control system of joining with the motherboard standard circuit.
To enter heat transfer model more convenient although Chatterbox server ranks make, but the line work of modulator-demodulator and network interface card has to carry out at the trailing unit of hot plug unit, like this, under the prerequisite of not mobile hot plug unit, it is very difficult that mobile line just becomes.And in traditional tower (tower) PC server, network interface card and nextport modem card NextPort insert the connector socket and enter the motherboard surface with vertical direction.In Chatterbox server ranks, these cards must be horizontally positioned in the motherboard surface in order to make the motherboard model vertically embed the mainboard frame.This just makes and enters netting twine and connect the difficulty that becomes, and changes circuit if desired and just has to move the motherboard module from the mainboard frame.
Some other suit bag has been applied to server.U.S. Patent number is that 5,877,938 server suit packet system provides heat exchange disc driver and power source model, and stacks the server board that slides according to dish on mainboard frame top with horizontal direction with a horizontal subdivision dish device.U.S. Patent number is that a layer frame dress computer processor model of 6,025,989 is arranged and disc driver, fan, motherboard and power supply can be arranged in order.
The suit bag of the multiple computer processor system of other model strives solving following problem, locatees and the multiple element of access by the back I/O mode partly of using a stage casing planar alignment effectively each element circuit to be divided into the front processor part and to separate from the plane, stage casing.In the supercomputer software kit architecture of having described transverse plane in the right angle in 5,251, No. 097 patent of the U.S., the I/O part forwardly is connected to the plane, stage casing, is connected to backboard at the rear portion.The U.S. 5,488, No. 541 patents have been described a VME backboard (this backboard has a plane, stage casing that can connect) a plurality of processors have been connected together.The U.S. 5,912, No. 802 patent has been described plane, a stage casing telecommunications interchanger rack that telephone service interchanger element is housed.
Can use the invention of different model motherboard to be awarded patent in the common tower device mode of PC, the United States Patent (USP) number be: 6,035,356 and 6,094,351.6,035, No. 356 patent is a crossover-platform structure for tower PC design, and this structure uses two boards to replace traditional single motherboard.First block of plate is a disposable plates, and second block of plate is the I/O plate.These plates are located with vertical direction, the terminal horizontal direction are placed in the space of an elongation and by pair of connectors with the skew terminal to connect.In this way, different motherboards can be connected on the I/O plate of a conventional design by pair of connectors.The U.S. 6,094, No. 351 patents: two different types of removable backboards can adapt to the motherboard of different model in common tower case.
Although making significant headway aspect computer processor, the server suit bag, high expandable internet super server still wishes a kind of device can be provided, and makes this server hold a large amount of commercial server mainboards in a rack that efficient electrical power and space arranged.
Summary of the invention
The present invention is the high expandable internet super server that is made of a plurality of commercial server mainboards, and these mainboards leave in the rack with enough electric power and space as the blade mainboard (blade) of server.These blade mainboards (blade) in the front portion of rack with vertical direction running fix.
The access platform (through plane) at mainboard frame (chassis) middle part provides shared electric power and peripheral control signal for all superserver blade mainboards (blade) in rack.The I/O signal of each superserver blade mainboard (blade) passes the access platform slit and arrives the interface card that is positioned at the rack rear portion.Because the existence of a common blade mainboard bearing structure (common blade carrier structure), the high scalability superserver can adapt to the server master board of different model in identical mainboard frame.The commercial motherboard of different model is installed in one and provides in the common blade mainboard bearing structure at unique mechanical interface for mainboard frame (chassis).A specially designed PCI mainboard can insert in the motherboard of different model, and this mainboard has two connectors, and first is connected to access platform, and second is connected to interface card.Redundant hot-swap efficient power is connected on the common power source signal on the channel plane.Mainboard comprises line management, and it can be assigned to electric power signal on the server flat board by imitating ATX electrical management agreement.It is following to be used for cooling off superserver that interchangeable fan tray is installed in blade mainboard (blade).Rack holds many row's superserver mainboards (blade), and these superserver mainboards (blade) and layer frame network switching and disk driver layer stack puts.
This high scalability superserver is compared with existing server in design has very big advantage.Be different from existing homogeneity server constructs and rely on the similar server flat board that same manufacturer provides, high scalability interference networks superserver is supported many commercial server flat boards that can be installed in the different model in the same mainboard frame (chassis).So just make HSP in same mainboard frame, use different server flat boards at different application.When these flat boards are in the state of being suitable for, just can guarantee the application of up-to-date ATX motherboard, and enjoy the advantage that ATX commodity motherboard increases reliability and reduces cost because of volume production.
Unique access platform setting provides the hot-swap ability of server blade mainboard (blade) and interface card.Do not need to take out interface card or network netting twine, individual blade mainboard (blade) just can be removed and replace.This just removes the maximum fault of safeguarding computer system, i.e. the problem that those are disconnected by the I/O netting twine or reclosing caused.The connection of all data paths all is an independent isolating, and the purpose of doing like this is in the process that blade mainboard (blade) is removed or inserted error in data to take place in order to prevent.In this framework, between second connector of mainboard and the interface card at access platform rear portion a passive I/O backboard is arranged.Transmit pci signal by the I/O backboard, a plurality of interface card of like this can row's formula each superserver mainboard being installed.
The mistake of different levels is efficiently revised in the design of high scalability superserver, and carries out redundancy replacement in a kind of mode efficiently between all elements.Server blade mainboard (blade) can be easy to and automatically be assigned to another client from a client.This makes mainboard is to respond the demand dynamic assignment blade mainboard that changes the client to become possibility.And allow HSP to utilize the residue server resource, otherwise these resources will be distributed to single client region statically.Unnecessary superserver blade mainboard (blade) can be distributed to a specific client and rapidly in the service of immediately reaching the standard grade when breaking down of certain server blade mainboard (blade).Power supply efficient and three times of redundancies preferably.Similarly, the use of multiple fan tray provides cold protection for each row's superserver blade mainboard (blade).Maintenance needs has been simplified in the global design of high scalability superserver.
First-selected framework adopts 19 inches NEBS racks, and an independent server cabinet can hold 32 superserver blade mainboards.This high scalability superserver compares with traditional 2U tier rack type dress server, and 75% the electric power and the demand of air-conditioning system have been saved in the space that provides identical server capacity but only to take half.
Still can keep efficient management and control outside the I/O network that the Circuit management of each superserver blade mainboard is all accomplished at super mainboard to the high scalability superserver.This circuit passes through to imitate standard A TX electrical management agreement to reach and the communication of the intermediary of motherboard processor, and report immediate status and erroneous condition are controlled the rising and the downward modulation of electric power.
When a plurality of racks were placed side by side, all superserver mainboards can connect and be controlled by an independent control desk and keyboard.Shared external equipment, for example keyboard, video, mouse, disc driver can be switched to any superserver mainboard selectively by the Systems Operator.
Description of drawings
Fig. 1 has showed the inside and outside structure (exploded view) of single cabinet-type high extending internet superserver with the visual angle of comprehensive solid.
Fig. 2 is four rack high extending internet superserver plane graphs placed side by side drawing from the angle of the place ahead observation.Fig. 3 is to the part observation of the mainboard frame of the super server in the Internet from the top.
Fig. 4 is to two inner three-dimensional explosion views of rack superserver placed side by side.
Fig. 5,6 and 7 be from top, the three-dimensional explosion views of fan tray of avris and full visual angle observation.
Fig. 8 and 9 is the stereograms from the redundant power of front and rear observation.
Figure 10 is the three-dimensional explosion views of the Yishanmen of high scalability superserver.
Figure 11-the 14th, the three-dimensional explosion views of the different installation step of superserver mainboard.
Figure 15 is the omnibearing observation of access platform.
Figure 16 is the place ahead observation effect of access platform.
Figure 17 is the stereogram that aerial lug is shared at the electric power supply system rear portion.
Figure 18,19 and 20 is omnibearing observation figure of traditional I/O backboard.
Figure 21-the 31st is for the better schematic diagram that embodies mainboard of the present invention.
Embodiment
With reference to Fig. 1, the overall view mapping of this high extending internet superserver 100 just presents particularly.For ease of describing, although we as you know the autgmentability of server allow a large amount of racks 110 to be assembled together to form an independent superserver 100, we describe earlier and are installed in the interior server 100 of rack 110.
With reference to Fig. 1-3, access platform 130 is separated front 112 in 114 in 110 racks from behind, and this part of rack is carried most of superserver mainboards 132 (blade) and most of interface card 134.Each server blade mainboard (blade) 132 comprises a processor at least, each interface card managing subsidiary disk storage associative processor at least one I/O information channel (for example, the array of a scsi disk, storage networking array (SAN) unit or a fibre channel (FC) disk array) or message exchange (for example FC switch, ATM switch, Ethernet etc. or modulator-demodulator).Qianmen flat board 124 is allowed 112 passages towards blade mainboard (blade) 132 from the front, and back door dull and stereotyped 124 allows 114 inlets to interface card 134 from behind.Even do not need the blade mainboard 132 on the mobile front 112, also be easy to from the rear portion 114 and enter and be used for the netting twine (not shown) of linkage interface card 134 and relevant magnetic disc store or message exchange equipment.Similar, we also can replace server blade mainboard (blade) 132 not needing to disconnect, disturb under the situation of these interface card 134 and netting twine.
Fig. 2 has embodied two mainboard frames 128 of separating preferably and has been placed on state in the rack 110 independently in overlapping mode, and each mainboard frame 128 provides the space for 16 superserver blade mainboards 132.In this drawing, each mainboard frame 128 comprises two stacked inferior mainboard frames (sub-chassis) 129, and each time mainboard frame 129 is that eight blade mainboards (blade) provide the space.What be about to describe below us is exactly by being installed in the complete mainboard frame 128 that a pair of power bay 144 under the mainboard frame 129 formed.In the structure that two mainboard frames 128 are arranged, one has 32 superserver blade mainboards 132 is contained in the single rack 110.Although have a pair of mainboard frame 129 and mainboard frame 128 relevant in above description and the every cabinet, we also can be understood as our present invention other arrangement mode of the superserver mainboard 132 of some, inferior mainboard frame 129 and mainboard frame 128 are explored.Also can understand like this,,, comprise that the arrangement pattern of frame-saw, rack or superserver mainboard 132 all is put within the category of the present invention about other layer frame of the present invention arrangement although the present invention has only described the situation that rack is equipped with multilayer mainboard frame 128.Blade mainboard of the present invention (blade) 132 is perpendicular positioning in each time mainboard frame 129, and other element also can be installed in the rack 110.
Shown in Fig. 2,3, most fan tray (fan tray) 140 all is movably mounted in the bottom of mainboard frame 128 with cooling server blade mainboard (blade) 132.With reference to Fig. 5,6 and 7, each fan tray 140 all has a plurality of individual fan units 142, and in the time of can guaranteeing like this to break down in certain fan unit, fan still can be made enough cold air with cooling superserver mainboard (blade) 132.There is one to insert the power supply connection at the rear portion of each fan tray 140.Fan tray is installed in time mainboard frame 128, and it can be fixed with the knob of manual manipulation ultra-large type screw or other form.In order to reach better effect, each time mainboard frame 128 has 3 fan tray 140, and 3 fan units 142 are arranged in each fan tray 140.
As Fig. 8, shown in 9, a large amount of power supplys 144 is installed in the power pack 146 in the mode that moves, and power pack 146 is installed in the mainboard frame 128 and separates with inferior mainboard frame 129.Power pack 146 comprises an AC/DC plate 148, and this plate adjustment also is sent to access platform 130 with this electric power from the electric power of power supply 144.In first-selected framework, have at least 3 removable power supplys 144 to be installed in the power pack 146, although in fact only need two power supplys just can provide power supply for whole assembling rack 110.This setup makes can be effectively alternative by redundant power when a power supply 144 breaks down.
Having four electric power controllers 144 is contained in the electric power control casket 146 in the mode of sliding by track 170.In first-selected framework, power supply 144 is 400 watts of high efficiency and multi-function out-put supplies, for example the HP3 series power supply of being produced by Magnetec company.The connector 172 that is positioned at AC/DC plate 148 front portions is additional on the connector 174 at power supply rear portion, makes the power supply can be by hot-swap.Front 178 parts of AC/DC plate 148 have the standard external connector 176 of a plug-in card program CD driver or hard disk drive.One cover is shared the rear portion that aerial lug 179 is installed in AC/DC plate 148.The direct aerial lug 304 of sharing aerial lug 179 and Figure 15 displaying has identical functions.Native system makes each mainboard 132 that this mainboard is connected with outside with the electronics high-speed switch.
As a kind of better implementation, AC/DC plate 148 equally also is included as the connector that shared external equipment is provided with, for example a CD driver 150 and a BOOT disc driver 152.The aerial lug that also promising other shared external equipment is provided with on AC/DC plate 148 or access platform 130 for example includes the general control platform of keyboard, mouse and display.
Show that as Fig. 2 each mainboard frame 128 comprises two mainboard frames 129, a power pack 146 and a ventilating fan pallet 140, height approximately be 37 ".When special requirement, in order to adapt to four processors are installed, the height of mainboard frame 128 approximately will reach 40 ".Fig. 1 has then embodied following characteristics from another point of view, there is an independent inferior mainboard frame (sub-chassis) can carry 8 blade mainboards 128 in each mainboard frame 128, places fan tray 140 and power pack 146 in the bottom of each time mainboard frame 129.
Exchanges data layer frame equipment 154, for example fiber channel data switching exchane or cross link data switching exchane also can be placed in the rack 110.In first-selected framework, the degree of depth of rack 110 and front and back all accesibility allow the before and after side in the same hole slot of rack that the 1U switch of two standards is installed.In another framework, layer frame disk storage unit 156 can replace one or more mainboard frames 128.Moreover because the degree of depth of rack suitably and from front and back can enter rack 110, in same hole slot, independent disk memory cell 156 can be doubled and redoubled according to the degree of depth of this element.
Tier rack type rack 110 can be realized the configuration of a lot of elements.The ancillary equipment of other model, the various shared peripherals of for example commutative memory device, different types of switch, modulator-demodulator etc. and picture floppy disk or disc driver etc. (as long as these elements directly or by an aggregated(particle) structure supporting layer frame installation cabinet) also can be installed in the mainboard frame 110.
With reference to Fig. 2,4, the high scalability superserver of forming by a plurality of racks 110 102 that will describe.In this system, two or more racks 110 are grouped together and constitute a high scalability superserver 102.Rack 110 is placed side by side, connects netting twine in order to make the tight and convenient adjacent rack of structure, need remove a side panel 120 of rack 110 at least.Although this method placed side by side has been saved the space effectively, in fact a plurality of racks 110 still can be separately independently mode discharge and line together.According to the layout of the machine room that high scalability superserver 102 is housed, the rack 110 of two or more rows can be regarded as the part of an independent high scalability superserver 102.Another kind of select rack 110 put a kind of structural arrangements mode that structure (outer circle arrangment) can be regarded single high scalability superserver 102 as with circle formula one or more or part, the rack that superserver blade mainboard 132 is housed is placed on the outer ring, and its front door 124 outwardly.DASD, SAN and network cabinet are placed on the center of outer ring.In a framework, can there be 1024 superserver blade mainboards (blade) 132 to form the part of high scalability superserver 102 at most, this superserver 102 is made up of four row's racks 110 (16 of every rows), and these racks 110 comprise the layer frame disk storage of 16 terabytes.According to the communication channel that is used to different mainboard frames are connected together, amount to 1000 racks 110 and can form a high scalability superserver 100, the electric power that this system provides is equivalent to 100,000 servers.Say that technically the size of a high scalability superserver 102 is not limited to use the technology of high scalability superserver 102 by the ground space decision.From practice, we expect that the big young pathbreaker of a high scalability superserver 102 is limited by the quantity of the rack that can arrange under the specific computer room environment.
In server 102, as replenishing of mode (side-by-sidearrangment) that rack 110 is arranged side by side, an isolating construction 160 (separation structure) is arranged in the middle of adjacent rack 110, and this isolating construction can provide a more easy inlet for netting twine and cooling air.As shown in Figure 4, the design of isolating construction 160 makes the profile of server 102 more attractive in appearance equally.In this structure, this isolating construction 160 comprises that one is taken over a business (topplate) 161, removable hub disk (removable rear plate) 162, a pair of mounting bracket (mounting brackets) 163 and lamp box (light box) and lamp prism (light bar lens) 165.A kind of better mode is that lamp box structure 164 is connected on a pair of fluorescent lamp in order to illuminate lamp prism 165.
Figure 10 has then showed the first-selected structure of Qianmen flat board 124.An outstanding hinge plate 180 is attached on the dull and stereotyped one side 181.Dull and stereotyped 181 by crooked and punching slightly, can guarantee circulation of air like this and make on as shown in Figure 2 the blade mainboard 132 indicating device as seen.Punching makes the profile of door-plate 124 become unusual equally.Take over a business 182 and chassis 183 be attached on dull and stereotyped 181 in order to more support to be provided.An outstanding bolt sheet (latch piece) 184 depends on the another side of flat board 181.A dull and stereotyped bolt sheet 185 is stabilized on the rack 110, if comprise that also the clockwork spring 186 that can strengthen elastic force is just better, like this, the keeper that comprises support 187 and button pin 188 just can be set.Identifier is attached on 189.
With reference to Figure 11-14, description be the better frame mode of superserver blade mainboard (blade) 132.Each superserver mainboard (blade) 132 comprises a motherboard (motherboard) 200, and this motherboard is contained in the shared mainboard filling structure 202 that unified mechanical interface is provided for rack 110.Mainboard filling structure 202 is sentenced removable mode in the front portion 112 of rack 110 and is located.Hereinafter will describe in detail, one group of connector 204 is located at the rear portion of each superserver mainboard 132.In these connectors 206,207, have at least the interface card connector 136 (Figure 16) on first and the interface card 134 to closely cooperate.Better mode is that second part in these connectors 208,209 will closely cooperate with access platform connector 300,302.(Figure 15)
The vertical direction setting of shared mainboard filling structure 202 makes it can adapt to the plane of mainboard 200, and this structure is deacclimatized largest mainboard 200 with its additional space, the space of reservation towards mainboard filling structure 202 front portions still arranged, in case of necessity to hold built in hardware driver or other built-in external equipments.In a kind of framework, the height of mainboard filling structure 202 is 12 ", width is 16.5 ", but in another kind of framework, in order to cooperate abaculus processor motherboard 200, the height of mainboard filling structure is 13.75 ", width is 16.5 ".
Each motherboard 200 all is the commercial motherboard that comprises a PCI connector 205 at least.Each superserver blade mainboard 132 comprises a mainboard 210 that is connected on the PCI connector 205, and usually, the plane of the plane of mainboard 210 and motherboard 200 is parallel.In this framework, mainboard 210 provides one group of connector 204 at its rear portion.Using a benefit that has the specially designed mainboard 210 of Standard PC I connector is to allow server master board 132 can adapt to a plurality of different commercial motherboards 200, these motherboards are placed in the mainboard filling structure 202, as long as the scale size on all these different motherboards 200 surfaces can be guaranteed with regard to not needing motherboard to be carried out any modification in its threading mainboard filling structure 202.Can comprise Pentium motherboard, SPARC ° motherboard and ALPHA ° of motherboard with server master board 132 shared commercial mainboards 200 of the present invention.Because the existence of this feature, no matter an independent server 100 is at same rack 110 or in same mainboard frame 128, can adapt to two different server master boards 132, these two mainboards have the different motherboards 200 of first and second different models respectively.And these two different motherboards can install different processors separately.
In this structure, the server blade mainboard (blade) 132 of three kinds of models is arranged here.The server master board 132 of first kind of model is optimized to a front-end server, and use this server website of supervising the cooking, and with consumer's interaction to respond inquiry.The server master board 132 of second kind of model is a database mainboard, and its effect is an information bank of seeking and excavate the client.The effect of the server master board 132 of the third model provides flow video and image.These three kinds of server master boards 132 can use the motherboard (motherboard) of various models.The application of best technique has been guaranteed in flexibility of the present invention, and need all technology is not provided or only be moved an operating system by single-vendor.The benefit that the present invention brought be except continuing the improvement technology aspect the motherboard, also be intended to can adapt to the variation on designing when some element such as bus, mainboard and I/O backboard have more new standard and technology.Figure 13 knows clearly and has shown a kind of better framework, mainboard bearing structure 202 comprises a base plate 220, this base plate 220 has a leading edge 222, a back edge 224 and an a pair of lateral edges 226,228, and pair of brackets 230 is attached to the back edge 224 of base plate 220 both sides base plates with screw or analog.Support 230 has determined the width of mainboard bearing structure 202, and this width is wideer than the distance between motherboard 200 planes and mainboard 210 planes, and is stretched over than the higher distance of the highest element on the mainboard.Header board 232 depends on a side of base plate 220 along the leading edge 222 usefulness screws of base plate 220.The width of header board 232 is by support 230 decisions, and this width is enough to compare with server master board bearing structure 202.A pair of traverse rod (Rails) 234,236 usefulness screws etc. depend on the back end of the anterior end of header board 232 and support 230.Closely cooperate along the perforation (projection) at a pair of edge 234,236 and pair of tracks 235 (see figure 3)s in time frame 129, for carrying rack 110 provides unified mechanical interface.Base plate 220 is metal preferably, and it is set in the insulator of rack 110 bottoms as electric interface.Being positioned between base plate (base plate) 220 and the motherboard (motherboard) 200 is insulating trip 242, the plane parallel of it and motherboard (motherboard) 200 and main card plate (host board) 210.As extra support, support 244 and a pair of perforation (projection) 246 are supported main card plate 210.Chi structure (A cross member) 248 makes the rear portion of blade mainboard (blade) bearing structure 202 more stable.
In the first-selected framework, header board 232 is equipped with a series of indicator light and button, and these indicator lights and button are on state of a control panel 250 and extend to a decorative plate 252 always.Assembling keeper 254 (latch assembly) scoops out with the loophole (projection) of keeper 256 and time frame 128 bottom trace, make connector 204 carry out machinery and be connected, and guarantee in the inferior frame 129 that server master board 132 is in detachable mode firmly in it with electric power with the connector of corresponding cooperation.On main card plate (host board) 210, the netting twine (not shown) links to each other state of a control panel 250 with connector.
With reference to Figure 15,16, be detailed description to access platform 130, access platform 130 is between the front 112 and back 116 of inferior frame 129.Preferred manner is, in each time frame 129 an independently access platform 130 arranged all, and in same frame 128 or rack 110, inferior frame 129 is with increase in demand or reduce and connect netting twine.Can understand like this, independent access platform 130 can provide supply for whole rack 110 or each frame 128.In better a kind of mode, at least two access platform connectors 300,302 provide supply for the server blade mainboard (blade) 132 that is contained in time frame 129, this frame and second portion 208,209 close fit that are positioned at one group of connector 204 of main card plate (host board) 210 rear part edges.Connector 300 comprises the electric power connecting line, and this connecting line and connector 208 cooperations are transferred to access platform 130 from AC/DC plate 148. Connector 302 and 209 cooperates, and comprises the line of the shared peripheral control signal of Connection Service device blade mainboard (blade).Connector 302 comprises a cover switch connection, share external equipment in order to guarantee server blade mainboard (blade) 132, for example control desk, startup disc driver and second cover uninterruptedly are connected, and make the operator directly will specify the periphery be connected on the particular server blade mainboard (blade) 132.
Figure 15 has shown the peripheral line 304 of a cover appointment, is installed in the rear portion of access platform 130 and permission is directly entered a special server blade mainboard (blade) 132 by the rear portion 114 of rack 110 outside connection.Upgrade electric power and aerial lug and do not need to replace access platform 130.Following mode is brought better effect, and connector 300,302 and 304 can be removed, and new connector depends on connection by precession access platform 130.
In access platform 130, have one and access platform connector 300,302 corresponding holes 306 at least.Hole 306 has been arranged, just can enter first part 206,207 of the connector group 204 that is positioned at main card plate 210 rear part edges from the rear portion of inferior frame 128.First part 206 of connector group 204 comprises the line of I/O signal, and this line is connected on the corresponding interface card of server blade mainboard 132, if one or more high density signal connector is for example just better from AMP.As what will describe in detail, better mode is to finish by independent I/O backboard 310, when server master board 134 is removed and the part 207 of connector group 204 when electric power being provided to I/O backboard 310 rear portion that allows interface card 134 to continue to insert I/O backboards 310.Other situation is, interface card 134 can be contained in the mechanical structure to guarantee that interface card 134 is in suitable position in 110 racks, and like this, server master board 132 just can directly insert in the interface card 134.But in this framework, the advantage with interface card 134 doublings that make each server blade mainboard 132 is not a nothing the matter.Here also have some arguments to approve of following saying, the I/O signal path that passes I/O backboard 310 has other advantage.
In the first-selected framework, access platform 130 by screw the weldering nail wait on the dull and stereotyped aggregated(particle) structure 320 of the middle-end that is fixed on a metal, 320 include and 310 corresponding holes 312, hole.The dull and stereotyped aggregated(particle) structure 320 of middle-end mechanically depends among inferior blade mainboard (blade) frame 129 (sub-chassis) in the frame 128.Because the existence of the dull and stereotyped aggregated(particle) structure 320 of access platform 132 and middle-end (midplane sub-chassis) is arranged, just can by electric power connect 330 with control sensing and be connected 332 and support and be connected fan tray 140.
With reference to Figure 18,19 and 20, description be I/O backboard 310.In this framework, I/O backboard 310 provides a connector 312, and the first 206 of the connector group 204 on this connector and the main card plate (host board) 210 is connected.The Standard PC I connector 314,316 of the rear portion some of I/O backboard 310 provides a plug-in type to connect, and is used to connect the independent standard interface card 134 of appropriate number.In the front portion of I/O backboard 310 an independent electric connector 318 is arranged, the first half in hole 306 that extends to access platform 130 is to be connected with electric connector 207 on the main card plate (host board).In this way, no matter be that server master board 132 or interface card 134 can disconnect connection by the electrical management circuit on the main card plate under the situation of not disturbing supply of electric power.
In this framework, send signals by I/O backboard 310, general pci signal, remove all non-mainstream signals (these non-mainstream signals make that each PCI slot is unique) after, from motherboard 200, be transferred in the PCI terminal by main card plate 210.The non-mainstream signal of each PCI slot (side-band signals) is all directly sent by main card plate 210 by the plate by low level.Arrive the rear portion high-density connector 206 of main card plate (host board) 210 up to signal, these lines all are equally long basically and send with rectilinear direction.The position of fastening on the main card plate 210 is connected on the high-density connector 206, should guarantee that pci signal cooperates the equal in length of target connector 314 and 316.The transmission of non-mainstream signal is finished, and the PCI slot of main card plate 210 will be sent on the high-density connector 206, be sent on high-density connector 206 lead-in wire farthest up to the PCI of the most close main card plate 210 slot.The isa bus signal is sent on the pci signal, and arrives I/O backboard 310 by access platform 130.
Although only the invention relates to the I/O backboard that designs for single hole 306, it still may make up an I/O backboard, and backboard runs through a plurality of holes 306 and passes through this big I/O backboard along an impartial verdict bus path allocation PCI and other I/O signal backboard.Main card plate and I/O backboard can use a PCI bridge, and the independent PCI slot of on the motherboard can be used to produce at the I/O backboard PCI signal of requirement like this, no matter on the motherboard (motherboard) how many PCI slots are arranged.
Although the invention relates to server blade mainboard 132 (blade), arrangement mode is 306 1 blade mainboards in each hole, it also may create server blade mainboard (blade) and blade mainboard (blade) bearing structure, and its width might occupy two or more holes 206.Server master board 132 can be made up of a plurality of processors, even each superserver blade mainboard (blade) is made of a plurality of motherboards, but they still keep the benefit of shared mainboard bearing structure and access platform.
With reference to Figure 21-31, the desirable combination of connector 204, mainboard 210 and I/O backboard 310 is described below, the signal that Figure 21 has described in the management processor 260 on main card plate 210, motherboard PCI connector 205 and the backboard signal connector 209 transmits.Figure 22 and 23 describes the pci bus signal in detail and transmits consistent with the principle of foregoing description with some are sent to other signal of I/O backboard 310 by connector 306 signal.How select from motherboard 200 (motherboard) by connector 205 Figure 24 describes pci signal in detail is.
Figure 25 has described peripheral peripheral signal and how to be sent to main card plate 210 (hostboard) up to connector 209 from motherboard 200.Figure 26 describes these peripheral peripheral signal in detail and how to change the foregoing peripheral signal I/O arrangement of sharing of execution.Figure 27 and 28 has described peripheral signal in detail and how to be sent to main card plate 210 up to connector 209 from motherboard 200.Figure 31 has described the management bus interface.
What Figure 29 represented is management processor 260 diagrams.The initiating sequence and the configuration information of 262 storages of lightning internal memory are used by management processor 260.Initiating sequence, configuration information and other control channel information are by a disjunct communication channel and the contact of main management processor, and this disjunct communication channel is counted as a management bus, by 264 management of the management bus interface on the main card plate 210.Circuit 266 collects voltage sensor information and these information is sent to preceding connector for substrate 267 and other some state informations show together.When interchanger 268 is shared peripheral peripheral signal in control, can manually select motherboard 200 and main card plate 210.Slot identifier 269 has been made the signal of a uniqueness for each superserver mainboard 132 in inferior mainboard frame (sub-chassis) 128.It is how to feel that level of power and ATX electric power connect 270 that drawing 30 is described circuit 266 in detail.Should understand like this, ATX electric power connects 270 and imitates the ATX initiating sequence, and this sequence is generally ATX motherboard 200 and exists.This initiating sequence is stored in the lightning internal memory 262, and like this, if when the agreement of ATX form key element motherboard 200 or standard have change, initiating sequence just can change.This design has utilized the advantage of agreement between each motherboard and its relevant ATX supply of electric power, microprocessor on the main card plate can use standard A TX communication initiating sequence to come the motherboard contact relevant with it, but still only allows industrial standard management software (for example LANsite) and motherboard contact.
Can start all server blade mainboards (blade) in a frame 128 simultaneously, the benefit that start-up routine is sorted is to make the possibility of power supply 144 overload drop to minimum.Because power supply 144 needs design and choose can deal with electric current and the voltage that initiating sequence causes and rise suddenly, the utilization initiating sequence just can be removed the required in this case excess power supply of doing from and design.Initial electric power offers server 100, and electric power is applied on all main card plates 210 simultaneously.Before startup ATX electric power sequence arrived relevant motherboard 200, each main card plate 210 (host board) had prolonged it and has fixed time earlier.Fixing time earlier is that whole set time numerical value multiply by the numerical value that a pumping slot identifier (269) derives out, and during startup, the selection of time numerical value is to have comprised shakiness of short duration when starting motherboard 200 to reach the standard grade up to next motherboard 200 and settle out.Selectively, each main card plate (host board) 210 can be waited for that receiving one up to it starts request by sequencing, this starts the start-up by hand of request from a supervising the network 264 or header board power switch 268, and this 268 switch provides electric power to act on server blade mainboard (blade) before for motherboard (motherboard).
Although we are described the details of first-selected framework, should be realized that still to have a large amount of variations and change here that scope of the present invention should be defined by the claims.
Claims (20)
1. high scalability superserver comprises:
The rack that the front and back is arranged; A large amount of server blade mainboards (Blade) are contained in the rack, each server blade mainboard comprises a motherboard, this mother matrix is at the anterior removable or perpendicular positioning of rack, and it first and second connector are located in the mode that can operate along the back edge of server blade mainboard;
An access platform is arranged between the front and rear of rack, and this access platform comprises: a large amount of access platform connectors, and these connectors exist for first connector with the server blade mainboard closely cooperates; There have been these holes in a large amount of holes, and connector just can the admission passage platform.A large amount of interface card is arranged in the rack, and each interface card is located with its plane vertical moving at the rear portion of rack, and the interface card connector closely cooperates by the part of second connector of corresponding hole and server blade mainboard in the access platform.
As claim 1 fast superserver, wherein, first connector comprises the peripheral signal that each server blade mainboard in the interface channel platform is common and the line of electric power, and second connector comprises and connect the line of I/O signal to corresponding blade mainboard interface card.
3. superserver as claimed in claim 2 wherein, further comprises the heat exchange supply of electric power device of a large amount of running fixes in rack, and this supply is connected on the line that electric power is provided in manipulable mode in access platform.
4. superserver as claimed in claim 3, wherein, motherboard is a commercial ATX motherboard, and it includes a PCI connector, each server blade mainboard comprises a main card plate that is operationally connected on the PCI slot again, and this server blade mainboard plane is parallel with the motherboard plane.The main card plate has two connectors and managing line at its rear part edge, and this managing line imitates ATX electrical management agreement to provide electric power to motherboard.
5. superserver as claimed in claim 1 also comprises a passive I/O backboard, and this backboard is between second connector in access platform rear portion and interface card.
6. superserver as claimed in claim 1, wherein, motherboard is a commercial motherboard that comprises the PCI connector, each server master board comprises a main card plate that is operationally connected on the PCI slot again, and two connectors are arranged at the rear portion of main card plate.
7. superserver as claimed in claim 1, wherein, each motherboard is installed in the shared mainboard bearing structure, this structure provides unified mechanical interface for the carrying rack, like this, in identical carrying rack, the high scalability superserver can adapt to the motherboard of different model.
8. superserver as claimed in claim 1 wherein, in the carrying rack, also comprises a large amount of fan tray, and the lower end that these fan tray are positioned at the server blade mainboard movably is used to cool off mainboard, comprises a large amount of fans in each fan tray.
9. according to claim 1, the rarest mainboard frame in rack comprises four server master boards at least in each mainboard frame, mainboard is installed in the superserver of a large amount of frames in frame, wherein, each time frame includes the server blade mainboard of arranging by row in a large number.
10. high expandable internet super server comprises:
A blade mainboard frame, this frame have a front and a back;
A large amount of superserver blade mainboards is contained in this frame, each server blade mainboard comprises a motherboard that is installed in the shared blade mainboard filling structure, this filling structure can provide a unified mechanical interface for the mainboard frame, the server bearing structure is in the anterior running fix of this frame, being operatively connected along the rear part edge of server blade mainboard is connector, a large amount of superserver blade mainboards comprise: first server blade mainboard, this mainboard contain first kind of model motherboard of subsidiary first kind of model processor; Second server blade mainboard, this mainboard contains the subsidiary second kind of model motherboard that is different from second kind of above model processor, a large amount of interface card are housed in the mainboard frame, each interface card also has an interface card connector to be used for closely cooperating with at least a portion of server blade mainboard connector here in the running fix of frame rear portion;
In same server blade mainboard frame, the high scalability superserver can adapt to the motherboard of different model.
11. superserver as claimed in claim 10, the server blade mainboard is located with vertical direction to the motherboard plane in the mainboard frame, and interface card is located with vertical direction by the interface card plane in the mainboard frame.
12. superserver as claimed in claim 10, the motherboard of each type is the commercial motherboard that comprises a PCI connector, each server blade mainboard comprises a main card plate that is operationally connected on the PCI connector, and each blade mainboard is to locate with the parallel plane direction of motherboard, and the rear part edge of main card plate is installed with connector.
13. a server master board that is contained in the superserver frame is made of following:
The blade mainboard filling structure of unified mechanical interface is provided for the blade mainboard frame; Different commercial motherboards is contained in the mainboard bearing structure, and these motherboards have a plane that has the PCI connector; Different motherboards has different planar dimension, and they all can be installed in the mainboard blade bearing structure; A main card plate is arranged in the blade mainboard bearing structure, and the operation of this main card plate and PCI connector links to each other and its plane and motherboard plane parallel; The connector that is positioned at this main card panel edges provides a unified electric power interface for superserver.
14. superserver as claimed in claim 13, wherein, the blade mainboard bearing structure comprises:
A base plate, this base plate comprise a leading edge, a back edge and a pair of lateral edges;
Pair of brackets, the operation of this support depends on the both sides of base plate near rear part edge, and this has determined the width of blade mainboard bearing structure to support, and this width is greater than the distance between motherboard plane and the main card board plane;
A header board, the side operation of this header board is attached to the forward edge of base plate, and has the width identical with the blade mainboard bearing structure; Here also exist a pair of traverse rod in addition, each traverse rod operation depends on the back end of an anterior terminal and support of header board.
15. superserver as claimed in claim 14, the base plate that comprises a loophole has defined a pair of edge, and this edge will think that the blade mainboard frame provides unified mechanical interface with a pair of track close fit of blade mainboard frame.
16. superserver as claimed in claim 14 wherein, is positioned at main card postlaminar part edge and has at least the part connector that motherboard and an interface card that has the PCI connector are operationally connected to together.
17. superserver as claimed in claim 14, wherein, base plate is a metal, and the server blade mainboard also comprises an insulating trip between base plate and motherboard, the plane parallel of this insulating trip and motherboard and main card plate.
18. a high scalability superserver system comprises:
Two sides before and after a large amount of blade mainboard frames, each frame all have;
In the blade mainboard frame, a large amount of server blade mainboards is arranged, each server master board comprises a motherboard that is installed in the shared mainboard bearing structure, this bearing structure provides unified mechanical interface to frame, and the mainboard bearing structure is in the anterior running fix of frame and comprise the connector that is positioned at the server master board rear portion;
A large amount of interface card is arranged in the blade mainboard frame, and each interface card is in the running fix of frame rear portion and have an interface card connector, and the minimum part of this connector and server blade mainboard connector closely cooperates;
The middle-end flat board that is installed between each blade mainboard frame front and back comprises a connector, and the server master board in the blade mainboard frame and the part of motherboard connector closely cooperate, for peripheral control signal provides connection; Have at least a signalling channel between the blade mainboard frame, to connect peripheral control signal.
19. superserver as claimed in claim 18 system, comprise a controlling platform, each server blade mainboard all is a management mainboard, manages the peripheral control signal of all other server master boards and the selectivity of at least a portion control desk periphery control signal and is connected.
20. superserver as claimed in claim 18 system, wherein, the middle-end flat board is an access platform, comprises a large amount of and the corresponding hole of blade mainboard connector, these holes have been arranged, just can enter a part with the close-fitting server blade motherboard connector of interface card.
Applications Claiming Priority (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US21860200P | 2000-07-17 | 2000-07-17 | |
| US60/218,602 | 2000-07-17 | ||
| US09/709,820 | 2000-11-10 | ||
| US09/709,820 US6452809B1 (en) | 2000-11-10 | 2000-11-10 | Scalable internet engine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1442032A true CN1442032A (en) | 2003-09-10 |
| CN1264078C CN1264078C (en) | 2006-07-12 |
Family
ID=26913077
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB018127932A Expired - Fee Related CN1264078C (en) | 2000-07-17 | 2001-06-28 | High expandable internet super server |
Country Status (8)
| Country | Link |
|---|---|
| EP (1) | EP1317876B1 (en) |
| JP (1) | JP2004519750A (en) |
| KR (1) | KR100859760B1 (en) |
| CN (1) | CN1264078C (en) |
| AT (1) | ATE520292T1 (en) |
| AU (1) | AU2001271563A1 (en) |
| CA (1) | CA2415769C (en) |
| WO (1) | WO2002007488A1 (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100403218C (en) * | 2005-05-24 | 2008-07-16 | 英业达股份有限公司 | Blade server system |
| CN100442198C (en) * | 2005-02-18 | 2008-12-10 | 蚬壳星盈科技有限公司 | Mounting design for blade type computer server and network exchanger and connecting method |
| CN101944667A (en) * | 2009-06-09 | 2011-01-12 | 泰科电子公司 | Be used for the connector flexible circuit member of electric coupling each other |
| CN101986239A (en) * | 2010-11-19 | 2011-03-16 | 曙光信息产业(北京)有限公司 | Rack-mounted blade system |
| CN102374182A (en) * | 2010-08-24 | 2012-03-14 | 联想(北京)有限公司 | Equipment cabinet and fan system thereof |
| CN102402256A (en) * | 2010-09-14 | 2012-04-04 | 鸿富锦精密工业（深圳）有限公司 | Server cabinet |
| CN102609049A (en) * | 2012-03-27 | 2012-07-25 | 中国人民解放军国防科学技术大学 | Sandwich back plate device for connecting plug-in cards |
| CN101719585B (en) * | 2009-11-24 | 2012-10-10 | 中兴通讯股份有限公司 | Network card and assembly method thereof |
| CN103324598A (en) * | 2013-07-17 | 2013-09-25 | 李斌 | Method for forming super computer by utilizing personal computers and internet |
| CN103869917A (en) * | 2012-12-12 | 2014-06-18 | 纬创资通股份有限公司 | Power supply module and server |
| CN104717866A (en) * | 2013-12-17 | 2015-06-17 | 国家电网公司 | Electric equipment mounting box frame |
| CN107371342A (en) * | 2016-05-12 | 2017-11-21 | 中兴通讯股份有限公司 | A kind of rack |
| CN108121411A (en) * | 2016-11-29 | 2018-06-05 | 英业达科技有限公司 | The server being easily assembled to |
| CN110716613A (en) * | 2019-09-14 | 2020-01-21 | 无锡江南计算技术研究所 | Integrated multi-section type operation plug-in interconnection assembly structure and method |
| CN112162608A (en) * | 2020-10-14 | 2021-01-01 | 上海万向区块链股份公司 | Hot plug server |
| CN113574754A (en) * | 2019-01-14 | 2021-10-29 | 美商新思科技有限公司 | Robotic system for joining server backplane connectors and corresponding method |
| CN115016615A (en) * | 2022-06-30 | 2022-09-06 | 西安易朴通讯技术有限公司 | PCIE card fixing device and case |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE10308869A1 (en) | 2003-02-28 | 2004-09-16 | Fujitsu Siemens Computers Gmbh | Optional slot for a blade server |
| US7136283B2 (en) * | 2003-06-11 | 2006-11-14 | Hewlett-Packard Development Company, L.P. | Multi-computer system |
| DE102008026538B4 (en) * | 2008-06-03 | 2010-05-27 | Fujitsu Siemens Computers Gmbh | Server system and for use in the server system suitable server and suitable connection module |
| TW201309179A (en) * | 2011-08-04 | 2013-02-16 | Wistron Corp | Airflow adjusting device and blade server |
| DE202019102256U1 (en) * | 2019-04-18 | 2020-07-21 | Dürkopp Adler AG | Carrying and docking device for electronics housing modules |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH06120636A (en) * | 1992-10-06 | 1994-04-28 | Oki Electric Ind Co Ltd | Interconnection structure for board |
| US5428806A (en) * | 1993-01-22 | 1995-06-27 | Pocrass; Alan L. | Computer networking system including central chassis with processor and input/output modules, remote transceivers, and communication links between the transceivers and input/output modules |
| WO1995024729A2 (en) * | 1994-03-11 | 1995-09-14 | The Panda Project | Modular architecture for high bandwidth computers |
| JPH07336074A (en) * | 1994-06-07 | 1995-12-22 | Pfu Ltd | Mounting structure of subunit for electronic apparatus |
| US5460441A (en) * | 1994-11-01 | 1995-10-24 | Compaq Computer Corporation | Rack-mounted computer apparatus |
| US5684671A (en) * | 1995-08-22 | 1997-11-04 | Sequent Computer Systems, Inc. | Packaging architecture for a data server |
| US5948087A (en) | 1996-12-16 | 1999-09-07 | Nexar Technologies, Inc. | Computer architecture having platform-dependent and platform-independent boards forming the motherboard |
| SG77151A1 (en) * | 1997-01-09 | 2000-12-19 | Sun Microsystems Inc | Full-featured special purpose network server |
| JPH1115580A (en) * | 1997-06-26 | 1999-01-22 | Hitachi Ltd | Rack mounting-type computer system |
| US5971804A (en) * | 1997-06-30 | 1999-10-26 | Emc Corporation | Backplane having strip transmission line ethernet bus |
| US6094351A (en) | 1998-08-19 | 2000-07-25 | Hon Hai Precision Ind. Co., Ltd. | Universal enclosure for different type mother boards |
| JPH11110083A (en) * | 1997-10-07 | 1999-04-23 | Hitachi Ltd | Information processor |
-
2001
- 2001-06-28 AU AU2001271563A patent/AU2001271563A1/en not_active Abandoned
- 2001-06-28 CA CA002415769A patent/CA2415769C/en not_active Expired - Fee Related
- 2001-06-28 JP JP2002513247A patent/JP2004519750A/en active Pending
- 2001-06-28 EP EP01950589A patent/EP1317876B1/en not_active Expired - Lifetime
- 2001-06-28 KR KR1020037000791A patent/KR100859760B1/en not_active IP Right Cessation
- 2001-06-28 WO PCT/US2001/020570 patent/WO2002007488A1/en active Application Filing
- 2001-06-28 CN CNB018127932A patent/CN1264078C/en not_active Expired - Fee Related
- 2001-06-28 AT AT01950589T patent/ATE520292T1/en not_active IP Right Cessation
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100442198C (en) * | 2005-02-18 | 2008-12-10 | 蚬壳星盈科技有限公司 | Mounting design for blade type computer server and network exchanger and connecting method |
| CN100403218C (en) * | 2005-05-24 | 2008-07-16 | 英业达股份有限公司 | Blade server system |
| CN101944667A (en) * | 2009-06-09 | 2011-01-12 | 泰科电子公司 | Be used for the connector flexible circuit member of electric coupling each other |
| CN101944667B (en) * | 2009-06-09 | 2014-03-12 | 泰科电子公司 | Flexible circuit member for electrically coupling connectors with one another |
| CN101719585B (en) * | 2009-11-24 | 2012-10-10 | 中兴通讯股份有限公司 | Network card and assembly method thereof |
| CN102374182A (en) * | 2010-08-24 | 2012-03-14 | 联想(北京)有限公司 | Equipment cabinet and fan system thereof |
| CN102374182B (en) * | 2010-08-24 | 2016-03-02 | 联想(北京)有限公司 | Rack and fan system thereof |
| CN102402256A (en) * | 2010-09-14 | 2012-04-04 | 鸿富锦精密工业（深圳）有限公司 | Server cabinet |
| CN102402256B (en) * | 2010-09-14 | 2016-09-14 | 鸿富锦精密工业（深圳）有限公司 | Machine box for server |
| CN101986239A (en) * | 2010-11-19 | 2011-03-16 | 曙光信息产业(北京)有限公司 | Rack-mounted blade system |
| CN102609049B (en) * | 2012-03-27 | 2014-06-04 | 中国人民解放军国防科学技术大学 | Sandwich back plate device for connecting plug-in cards |
| CN102609049A (en) * | 2012-03-27 | 2012-07-25 | 中国人民解放军国防科学技术大学 | Sandwich back plate device for connecting plug-in cards |
| CN103869917A (en) * | 2012-12-12 | 2014-06-18 | 纬创资通股份有限公司 | Power supply module and server |
| CN103869917B (en) * | 2012-12-12 | 2017-05-10 | 纬创资通股份有限公司 | Power supply module and server |
| CN103324598A (en) * | 2013-07-17 | 2013-09-25 | 李斌 | Method for forming super computer by utilizing personal computers and internet |
| CN104717866A (en) * | 2013-12-17 | 2015-06-17 | 国家电网公司 | Electric equipment mounting box frame |
| CN107371342A (en) * | 2016-05-12 | 2017-11-21 | 中兴通讯股份有限公司 | A kind of rack |
| CN108121411A (en) * | 2016-11-29 | 2018-06-05 | 英业达科技有限公司 | The server being easily assembled to |
| CN108121411B (en) * | 2016-11-29 | 2021-03-12 | 英业达科技有限公司 | Easy-to-assemble server |
| CN113574754A (en) * | 2019-01-14 | 2021-10-29 | 美商新思科技有限公司 | Robotic system for joining server backplane connectors and corresponding method |
| CN110716613A (en) * | 2019-09-14 | 2020-01-21 | 无锡江南计算技术研究所 | Integrated multi-section type operation plug-in interconnection assembly structure and method |
| CN112162608A (en) * | 2020-10-14 | 2021-01-01 | 上海万向区块链股份公司 | Hot plug server |
| CN115016615A (en) * | 2022-06-30 | 2022-09-06 | 西安易朴通讯技术有限公司 | PCIE card fixing device and case |
Also Published As
| Publication number | Publication date |
|---|---|
| JP2004519750A (en) | 2004-07-02 |
| KR100859760B1 (en) | 2008-09-24 |
| EP1317876A4 (en) | 2009-01-07 |
| KR20030066586A (en) | 2003-08-09 |
| ATE520292T1 (en) | 2011-08-15 |
| AU2001271563A1 (en) | 2002-01-30 |
| EP1317876A1 (en) | 2003-06-11 |
| WO2002007488A1 (en) | 2002-01-24 |
| CA2415769C (en) | 2005-08-30 |
| EP1317876B1 (en) | 2011-08-10 |
| CA2415769A1 (en) | 2002-01-24 |
| CN1264078C (en) | 2006-07-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1442032A (en) | High expandable internet super server | |
| Barroso et al. | The datacenter as a computer: An introduction to the design of warehouse-scale machines | |
| US6452809B1 (en) | Scalable internet engine | |
| US6583989B1 (en) | Computer system | |
| US9811127B2 (en) | Twin server blades for high-density clustered computer system | |
| CN2852260Y (en) | Server | |
| US6510050B1 (en) | High density packaging for multi-disk systems | |
| TW563033B (en) | Modular server architecture | |
| CN100442198C (en) | Mounting design for blade type computer server and network exchanger and connecting method | |
| US20100008038A1 (en) | Apparatus and Method for Reliable and Efficient Computing Based on Separating Computing Modules From Components With Moving Parts | |
| CN1503946A (en) | Server array hardware architecture and system | |
| US20130077223A1 (en) | Server | |
| CN1643475A (en) | Modular server system structure of ethernet route via support plane using integrating ethernet switching module | |
| CN101639712A (en) | Server | |
| US20170112015A1 (en) | Server | |
| Barroso et al. | The datacenter as a computer | |
| US8312144B2 (en) | Method, apparatus, and system for exchanging services in a distributed system | |
| CN103034302B (en) | Servomechanism | |
| US20100027214A1 (en) | Motherboard module array | |
| CN109976465A (en) | Server | |
| GB2356748A (en) | Redundant power supply for multiple processor system | |
| CN115481068B (en) | Server and data center | |
| EP1146430A2 (en) | Card design having tape and disk drives | |
| Watts et al. | Implementing an IBM system x iDataPlex solution | |
| CN215416537U (en) | Multi-node hot plug 4U storage server |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20060712 Termination date: 20180628 |